Re: [eigen] GitLab migration is starting now!

[ Thread Index | Date Index | More Archives ]

We used to have a linear-history policy when we started with mercurial -- obviously mistakes where made over time (and I'm definitely not happy with that when browsing through the history, doing bisects, etc)

At least for trivial changes all relevant information should be contained in the commit-message itself (of course one question is until when is a change "trivial")

I found the setting (under "Settings/General/Merge requests"). There is also a semi-linear history option (essentially the same, but with a merge-commit), which might be a compromise to at least avoid commit graphs spanning over hundreds of commits).

Especially, once we seriously start with CI, I totally agree that having all relevant information of the PR available would be great.

Other opinions?


On 05/12/2019 19.15, Joel Holdsworth wrote:
That's not typical usage in GitLab.

The main reason not to is that it will destroy the information from the MR.

If you do a merge, it will include any text I write in the MR description, the user who submitted the MR, the comitter, the MR's ID etc. - which is all useful information for record keeping.

If you cherry-pick my commits you will lose all that.

It would make sense to keep the commit history if it were linear from the start like the Wine project's Git repo, but for history which already has many merges, I don't see much benefit, personally.


On 12/5/19 6:07 PM, Christoph Hertzberg wrote:
I don't seem to be able to rebase/fast-forward merge this (I don't like polluting our history with trivial changes).

@Gael: Are there any project settings which need to be done for that?


On 05/12/2019 18.57, Joel Holdsworth wrote:
Ok - it's working now. Here is MR #1 !

Does it deserve such a momentous designation? You be the judge.


On 12/5/19 5:27 PM, Christoph Hertzberg wrote:
I should have checked for new mails, before sending my last.

I just granted you Reporter rights (as well as to two other people who requested access -- I did not grant Developer access, of course).

I think the manual approval of new members won't be as tedious as the previous procedure of granting access to our bugzilla. And having a manual step with a small time delay might be a good way to avoid bug/PR-SPAM.


On 05/12/2019 18.16, Joel Holdsworth wrote:
Ok... manual approval sounds tedious. I would recommend figuring out a way to automate the approval e.g. with a button on the website. Otherwise, it's just pointless admin for you guys.

Anyway, my username is "jhol", so can you make me reporter?


On 12/5/19 5:07 PM, Gael Guennebaud wrote:

On Thu, Dec 5, 2019 at 5:33 PM Joel Holdsworth <joel@xxxxxxxxxxxxxxxxxxx <mailto:joel@xxxxxxxxxxxxxxxxxxx>> wrote:

    Hi Gael!

    Thanks for this. It's going to make my work so much easier.

    I just wanted to resubmit my int8 ostream patches, but it seems that I     dodn't have permissions to submit a merge request. Do you need to open
    up the MRs to non-members?

thank you for trying. I was indeed unsure whether MR was open to anybody. According to this page:, MR is open for "reporter" members only, meaning that we'll have to manually approve people wanting to join the project to propose MR.

For the record, compared to "guests" (aka everybody for a public project as Eigen), reporters have the following additional permissions:

View Security reports
See a list of jobs
See a job log
Download and browse job artifacts
View confidential issues
Assign issues
Label issues
Lock issue threads
Manage issue tracker
Manage related issues
Manage labels
Create code snippets
See a commit status
See a container registry
See environments
See a list of merge requests
View project statistics
View Error Tracking list

This seems OK to me.



    On 12/4/19 9:17 PM, Gael Guennebaud wrote:
     > The critical steps of the migration are done. I've thus made
    public the
     > new repo:
     > I've also re-opened the deprecated mercurial repo on bitbucket so
     > people can smoothly migrate their clones, links, etc.
     > Please report any outdated links and mercurial/bitbucket references      > you'll find. I'm already aware of some pages on our wiki, like:
     > -
     > -
     > Any help on updating those old documentations will be greatly
     > Gael
     > On Wed, Dec 4, 2019 at 9:04 AM Gael Guennebaud
     > <gael.guennebaud@xxxxxxxxx <mailto:gael.guennebaud@xxxxxxxxx>
    <mailto:gael.guennebaud@xxxxxxxxx>>> wrote:
     >     Hi,
     >     as planed we'll start the migration to GitLab today. This
    means that
     >     the Eigen's bitbucket project (repository+pull-requests) won't be      >     accessible within a few minutes. Same for our bugzilla that
    will be
     >     turned as read-only.
     >     Wish me good luck and see you soon for announcing the
     >     of the new git repository!
     >     Cheers,
     >     Gael

 Dr.-Ing. Christoph Hertzberg

 Besuchsadresse der Nebengeschäftsstelle:
 Robotics Innovation Center
 Robert-Hooke-Straße 5
 28359 Bremen, Germany

 Postadresse der Hauptgeschäftsstelle Standort Bremen:
 Robotics Innovation Center
 Robert-Hooke-Straße 1
 28359 Bremen, Germany

 Tel.:     +49 421 178 45-4021
 Zentrale: +49 421 178 45-0
 E-Mail:   christoph.hertzberg@xxxxxxx

 Weitere Informationen:
  Deutsches Forschungszentrum für Künstliche Intelligenz GmbH
  Trippstadter Straße 122, D-67663 Kaiserslautern, Germany

  Prof. Dr. Antonio Krüger (Vorsitzender)
  Dr. Walter Olthoff

  Vorsitzender des Aufsichtsrats:
  Dr. Gabriël Clemens
  Amtsgericht Kaiserslautern, HRB 2313

Mail converted by MHonArc 2.6.19+